Background: The main strength of randomised controlled trials is based on each group being balanced in all prognostic characteristics. Intention to treat (ITT) analysis helps to preserve such balance and it should be the main approach for analysis in randomised controlled trials. Post-randomisation exclusion of patients may prevent proper ITT analysis and introduce bias. Some authors attempt to overcome the problem of post-randomisation exclusion by introducing the 'modified intention to treat' (mITT) analysis which according to our former analysis allows an arbitrary patient inclusion criteria.
Objectives: The aim of this study is to highlight the methodological quality of randomised controlled trials using mITT with respect to trials using a true ITT analysis. Specific focus will be given to the handling of post-randomisation exclusions.
Methods: We performed a computerised search in Medline and Cochrane controlled trials register to retrieve randomised controlled trials published in 2005 and 2006 in three major medical journals and other three specialised journals that were more likely to report trials with the mITT approach. Each trial was evaluated by two independent reviewers. The following data were extracted from the articles: characteristics of the trials including concealment of allocation and appropriate blinding, number of patients, number of treatment arms, intervention studied, use of placebo, calculation of sample size; the approach used for primary or secondary analysis: ITT, mITT, per-protocol; and data related to deviation from random allocation, false inclusion, missing response and the handling of missing data. Methodologic quality was compared between trials reporting the mITT and true ITT approach.
Results and conclusions: Randomised trials published in JAMA, Lancet, NEJM, Antimicrobial Agents and Chemotherapy, American Heart Journal and Journal of Clinical Oncology were considered for evaluation. Results will be presented at the Colloquium.